Mesa, AZ. The Internet of Things was a hot topic at the Burn-in and Test Strategies Workshop this week, despite some caution on the subject voiced by Risto Puhakka, president of VLSIresearch, in a Sunday evening address.
But in technical session Monday morning, presenters dove right into the topic. Laurie Wright of Xcerra discussed how the IoT will change back-end processing. Anthony Lum of Advantest commented on how to contend with high-mix, low- to moderate- volume devices. And Robert Howell from Exatron described serialization techniques for high-volume IoT security.
In a paper titled “How Internet of Things Will Change Back End Processing,” Wright, director of business development, related technology requirements and business challenges. IoT markets, she said, extend from fitness and wellness to smart cities, and IoT devices must be smart, low power, connected, and secure.
“IoT is a dynamic and evolving market,” she said. “Every day new applications coming out.” Unfortunately, she added not all are high volume—in fact, the majority may be moderate and low volume. And they will all need fast development time.
She divided the connected world into three areas—the edge node, of which there will be billions; edge-to-cloud aggregation devices, of which there will be millions, and data-crunching services, which will account for thousands of devices.
IoT devices will include digital, analog, and RF circuitry as well as embedded memory, and for application like automotive, they will have power capabilities. Wright said they will require transaction-based (protocol-aware) test-pattern methodologies and scalable and reconfigurable testers that can address a high mix of IoT devices. Multisite will likely be a valuable approach, but Wright noted requirements for site-dependent data, such as MAC addresses and calibration information.
In addition, IoT devices present handler challenges, she said. The handlers must provide physical stimulus—such as acceleration, pressure, gas, temperature, light, sound, and touch—to sensors. And the devices will come in a variety of package types (such as WLCSP) and sizes (less than 1 mm x 1 mm).
“Handling and test equipment will need to be flexible and adapt to the technical and business challenges of the dynamic IoT market,” she concluded.
Lum, business development manager at Advantest, echoed many of Wright’s points in his paper titled “Internet of Things Testing Challenges.” He agreed, for example, that the IoT will present high-mix test scenarios.
He cited some examples of IoT products, including GoPro Hero4, Google Contact Lens (which monitors blood sugar level), a wireless bridge sensor, a pacemaker that communicates with a smartphone, and even a wireless diaper.
He also divided IoT devices into three categories: small (personal and medical sensors, for example), medium (cellphones and automotive devices), and large (communications infrastructure and server components).
For small devices, he said, low cost test is critical. He presented two solutions to the cost-of-test question: higher site counts on existing ATE or lower cost test hardware. For the former, he said, multisite efficiencies (MSE) will need to be high for the techniques to be effective. As for the latter, the EVA (Evolutionary Analog) modular test system can offer a low cost alternative to more expensive testers like the T2000 and V93000.
As for medium the medium category, cellphone devices will have more cores, more sensors, more RF capabilities, and more security features. Automotive devices will require more scan vectors as well as more analog and RF tests—all over temperature. Scan depths, he said, are expected to double every three years.
He also predicted more concurrent test—concurrent logic, then concurrent memory, concurrent calculations (DSP), and ultimately concurrent analog/RF.
As for infrastructure, he said Internet traffic will exceed 88.4 EB per month in 2016 and double by 2019. The test of infrastructure and related devices can benefit from stacked-memory test methods now under development, but final-assembly memory test will remain a concern. In addition, it will be necessary to develop optical-signaling tests and fixturing.
In conclusion, Lum said, “IoT test challenges require IoT-specific solutions. Many challenges have been solved—some remain. We’re working hard on those.”
Also in the session, Howell, president and CEO of Exatron, offered a presentation titled “Serialized Programming Solutions for IoT Secure Elements.” Smart technology is on the rise, he said. Smart devices must be paired with consumers’ phones and tablets, and EMV card readers must be paired with bank software. “Fraud is easily detected if serial numbers are unique and paired,” he said.
But serialization imposes challenges, he said: “One incorrectly programmed device can waste a production lot of thousands.” He described an automated approach to programming serial numbers both internally and in human-readable form externally.
“The future of smart technology depends largely on security,” he concluded. “Security depends largely on trackable serial numbers. Automation guarantees perfect serialized output, logged and tracked at each production step.”
